Shane Loretz

> The problem with POD types is that they can't be extended without breaking ABI. Breaking ABI is only a problem if the the POD structs are part of the...

Seems like this is the same issue as #90

I think this can be fixed in `melodic`. The downstream typedefs are identical in https://github.com/ros-controls/realtime_tools/commit/b5c6cc9383a6855e33040e6d41bddcdc7b89282c which [is released into melodic](https://github.com/ros/rosdistro/blob/680852343e8daf674034b35f3f8f23f37704c6d6/melodic/distribution.yaml#L5505). The compiler error in the original report should be gone...

> do you have suggestions for exemplary ament_cmake packages containing python extensions? `rclpy` and `rosbag2_py` are the only two I can think of that create CPython extensions. IIUC they both...

Is this related to #20, which was fixed by #21?

> Should this not be fixed in [mav_msgs](https://github.com/ethz-asl/mav_comm/tree/master/mav_msgs)? @ffurrer I don't think so. The library target is named `mav_msgs`, but does not come from `mav_comm/mav_msgs`. It is created here: https://github.com/ethz-asl/rotors_simulator/blob/748268f9a11f6473cfeb55827459cea1e1d7b2d5/rotors_gazebo_plugins/CMakeLists.txt#L205-L212

> If its installed from source I can't really reproduce the problem (were having plenty of people running melodic and rotors) @michaelpantic I don't have a workspace at the moment,...

Friendly poke, are there any changes that need to be made to this PR before it can be merged?

Friendly ping :)

friendly ping :)